Comparison between computed tomography and ultrasound for presurgical evaluation of oral tongue squamous cell carcinoma tumor thickness

Abstract

Purpose:

To compare the accuracy of oral tongue squamous cell carcinoma (OTSCC) tumor thickness (TT) measured on CT to intraoperative ultrasound (US) and histopathology.

Methods and materials:

Twenty-six patients with OTSCC who underwent tumor resection by a single surgeon with simultaneous intraoperative US between 3/2016 and 4/2019 were prospectively identified, and their data reviewed. TT was independently measured in 19 patients who underwent preoperative CT (cTT) by two neuroradiologists blinded to US and histological results. The confidence level of interpretation of cTT was recorded by each reader using a 5-point Likert scale. The degree of dental artifact on CT was also scored. cTT was compared to TT measured on intraoperative US (uTT) and histopathologic assessment of TT (hTT).

Results:

OTSCC was visualized on CT in 52% (10/19) and 63% (12/19) of cases for readers 1 and 2, respectively. Mean Likert score was 0.42 for reader 1 and 0.73 for reader 2. Mean cTT of OTSCCs was 5.8 mm +/− 1.7 mm (n = 11). In comparison, mean uTT and hTT were 7.6 mm±3.5 mm and 7.1 +/− 4.2 mm, respectively. The Pearson coefficient (95% confidence interval) was 0.10 (−0.53–0.66) between cTT and hTT (n = 11) and 0.93 (0.74–0.98) between uTT and hTT.

Conclusions:

Preoperative CT is not reliable for assessment of TT in OTSCC compared to US and histopathology, particularly for OTSCC under 10 mm. US offers a practical complementary imaging tool with a unique role for primary tumor assessment that can aid in pre-operative planning, especially for small tumors.

1. Background and purpose

Accurate assessment of tumor thickness (TT) and depth of invasion (DOI) is critical for the clinical staging and management of oral tongue squamous cell carcinomas (OTSCC) [1]. TT represents the thickness of a tumor from its superficial surface to the deepest point of invasion, while maximum DOI refers to the thickness of a tumor from the level of the basement membrane to the deepest point of tumor invasion on a linear plane. Preoperative evaluation of tumor extent primarily involves CT, MRI, and/or manual palpation. It is generally accepted that the utility of CT is to assess for deep invasion beyond the oral tongue, for example to the floor of mouth or mandibular marrow, and not so much for assessment of small tumors confined to the oral tongue. However, there is insufficient data on the actual performance of CT for assessment of smaller tumors that are confined to the oral tongue and on whether suitable supplementary imaging may be used for more comprehensive pre-operative planning for tumors of all sizes. At present, ultrasound (US) is not routinely used in preoperative assessment, although it is emerging as a reliable and accurate modality for OTSCC assessment [2]. In this study, we examined the reliability and accuracy of preoperative CT in assessing TT of OTSCC (cTT) with those from intraoperative US (uTT) and histopathologic assessment (hTT). Of note, the recently updated, eighth edition of the Cancer Staging Manual of the American Joint Committee on Cancer now considers DOI to be an independent, key measurement for determination of the T category in tumor staging [1]. However, DOI is not readily assessed by CT since the tongue is often closely apposed to adjacent surfaces such as the palate or buccal surface, and the natural contour of the tongue surface and thus the basement membrane cannot be readily discerned. As such, for this study, we focus our attention to TT only.

2. Methods

2.1. Patient demographics and clinical data collection

This was an IRB-approved study performed at an integrated hospital system within the northeastern USA. Patients were identified for inclusion into this study if they a) had a diagnosis of a new/previously untreated OTSCC based on clinical examination findings and tissue biopsy, b) had a preoperative contrast-enhanced neck CT within our hospital network, c) underwent intraoperative US of the tongue during tumor resection, and d) had histopathologic analysis of the resected oral tongue tumor where the tumor thickness was recorded (hTT). All surgical procedures in this patient cohort were performed by a single head and neck surgeon with over 30 years of experience (M.A.V). Exclusion criteria included patients who had significant motion artifact precluding a diagnostic assessment on either US or CT imaging.

Electronic medical records were reviewed on all patients included into this study. Basic demographic information including age at the time of diagnosis of oral tongue squamous cell carcinoma and gender were recorded. In addition, the tumor laterality (right versus left) was recorded based on clinical examination information and hTT was recorded from histopathology records of our subspecialized system.

2.2. CT Technique and image analysis

Contrast-enhanced CT examination (CECT) of the neck was performed as either a dedicated CECT of the neck performed in isolation (n = 15) or in conjunction with a PET examination as a diagnostic CECT (n = 4) (as opposed to a non-CECT for attenuation correction).

All CT examinations were evaluated by two readers, both dedicated neuroradiologists, one with greater than 3 years and one with greater than 5 years of neuroradiology experience (B.C.Y and K.B). The readers performed independent analyses of all CT scans in the patient cohort. The readers were blinded to all clinical information regarding the patient and OTSCC other than tumor laterality (left side versus right side) which was provided from the clinical note. The decision was made to let the readers be aware of the side of the oral tongue lesion, since this mirrors real-world situations when this piece of clinical information is generally available to radiologists during study interpretation at the workstation in routine clinical practice.

The CT imaging appearance of OTSCC was defined as a relatively hyperdense lesion, compared to the attenuation of the remaining mucosa and musculature of the oral tongue, present on at least two-contiguous CT scan slices (Fig. 1A).

Fig. 1.

27-year-old male with a right-sided OTSCC. A: Axial CT image at the level of the oral tongue shows a lentiform, hyperattenuating lesion involving the right lateral aspect of the oral tongue (white arrow), which correlates to the site of a known OTSCC. B: Corresponding coronal CT image through the OTSCC (white arrow), with cTT measurement denoted by the double headed white arrow. Note, this patient would be considered to have mild dental artifact. Scale bar = 10 mm.

Measurements of the OTSCC were performed on MPR reformats created on either an Advantage 4.6 workstation (GE Healthcare; Chicago, IL, USA) or through a Visage 7 viewer (Visage Imaging, Berlin, Germany). All MPR reformatted images were created by defining a long axis plane parallel to the long axis of the OTSCC. The short axis of the OTSCC was then measured on the coronal MPR as the axis perpendicular to the tangent drawn along the long axis of the tumor image. This short axis measurement was designated as CT tumor thickness (cTT) (Fig. 1B).

Each reader independently created the MPR reformats and recorded short axis measurements of the OTSCC at the time of data analysis. Each reader also recorded his/her level of confidence for the detection of the OTSCC based on a 5-point Likert scale with 0 = not at all confident; 1 = not confident; 2 = neutral; 3 = somewhat confident; 4 = very confident. CT cases with discrepant recordings between the two primary readers was resolved through adjudication with a third reader, an experience head and neck radiologist with greater than 16 years of experience reading head and neck studies (M.B.C).

Dental restoration material was noted to generate significant artifacts through the oral cavity (Fig. 2) and limited assessment of the oral tongue in several cases. A consensus agreement was made determining the degree of streak artifact generated from dental restoration material, and was scored as none, mild, moderate, and severe. The degree of streak artifact from dental restoration material was recorded for each patient.

Fig. 2.

50-year-old female with OTSCC on the right side on physical exam. A: Axial CT image through the oral tongue. B: Coronal CT image of the oral tongue. Severe dental artifact is seen on both images A and B, limiting evaluation of the oral tongue. The known OTSCC was not visualized on CT in this case. Scale bar = 10 mm.

2.3. Intra-operative ultrasound technique

All patients underwent an intra-operative US of the oral tongue between 3/31/2016 and 4/26/2019. All US examinations were performed on a Philips Healthcare platform with a broadband compact linear array transducer (15-7io; Philips Healthcare) utilizing the imaging technique previously described in the study by Yoon et al. [2]. Intra-operative US examination was performed and evaluated by a single dedicated head and neck radiologist with greater than 10 years of experience in head and neck radiology (A.F.J). US images were acquired in both the longitudinal and transverse planes with respect to the long axis of the tumor.

2.4. Intra-operative ultrasound analysis

The tumor thickness identified from uTT was recorded at the time of interpretation of the US images by 2 neuroradiologists (A.F.J and B.C.Y). On PACS, the uTT was measured from the superficial surface of the tumor to the deepest margin of tumor invasion perpendicular to the plane of the tongue surface [2]. Consensus measurements between both readers were used for analysis.

2.5. Statistical analysis

The mean and standard deviation for the CT confidence scores were calculated for each reader. The correlation between the measurements from cTT and hTT as well as from uTT and hTT was assessed by calculating the Pearson coefficient. The statistical analysis was performed using GraphPad Prism software (Version 8).

3. Results

Of the 26 patients who underwent glossectomy, 19 patients had undergone preoperative CT with intravenous contrast. The mean age of the 19 patients was 59.6 years old (range = 27–82 years old) and there were 13 males and 6 females.

OTSCC was visualized on CT in 52% (10/19) and 63% (12/19) of the cases for readers 1 and 2, respectively. Nine patients had tumors visualized by both raters with a mean cTT (+/− standard deviation) of 6.8 mm +/− 1.7 mm for reader #1 and 5.2 mm +/− 1.7 mm for reader 2.

The overall mean cTT of the tumors was 5.8 mm +/− 1.7 mm with a range of 3 mm – 9 mm (n = 11). In comparison, the mean uTT and hTT were 7.6 mm +/− 3.5 mm (range of 4 mm – 15 mm) and 7.1 +/− 4.2 mm (2.5 mm – 15 mm), respectively. The median tumor thickness also showed a similar discrepancy with cTT of 5 mm, uTT of 8 mm, and hTT of 9 mm. On histopathology, OTSCC tumor thickness was on average 7.1 mm (range 2–16 mm) with only 2/11 (18%) tumors measuring greater than 10 mm in thickness.

Comparison between TT measurements from CT and histopathology was poor with a low Pearson coefficient (95% confidence interval) of 0.10 (−0.53–0.66; n = 11; Fig. 3). In comparison, the TT measurements from US and histopathology showed an excellent correlation with a Pearson coefficient of 0.93 (0.74–0.98; n = 11; Fig. 3). Tumor thickness as assessed on CT, US, and histopathology is depicted for each of the 19 patients (Fig. 4).

Fig. 3.

Correlation of TT measurements between different modalities. Left: the correlation between tumor thickness as assessed on CT (cTT) is compared to the tumor thickness as assessed on histopathology (hTT). The correlation between cTT and hTT was poor with a Pearson coefficient (r) of 0.10. Right: the correlation between tumor thickness as assessed by ultrasound (uTT) is compared to hTT, which yielded an excellent correlation with r = 0.93.

Fig. 4.

Histogram plotting tumor thickness as assessed on CT (cTT), ultrasound (uTT) and histopathology (hTT) for each of the 19 patients in this cohort. Where there is no recorded cTT, tumor was not visible to the readers on CT.

CT diagnostic confidence scores were recorded by each reader for all 19 cases, with a mean of 0.42 for reader 1 and 0.73 for reader 2 (range for both readers 0–4), denoting that on average both readers felt ‘not confident’ for the CT assessment of OTSCC.

Artifact from dental hardware and restoration material was noted to limit evaluation of the oral tongue for many cases. This artifact was deemed severe in 31% (6), moderate in 42% (8), and mild in 21% (4) of the cases. One case (5%) did not have any artifact from dental restoration material. For 5/6 (83%) of cases with severe dental artifact, the known OTSCC was not visible to either reader. One case with severe dental artifact had an OTSCC which was identified by both readers and was the largest tumor in the cohort measuring 16 mm in thickness on histopathology.

4. Discussion

The results of this study demonstrate that CT measurements for OTSCC tumor thickness were neither reliable nor accurate compared to histopathologic measurements. Measurements of tumor thickness assessed on US had a much higher (‘excellent’) degree of correlation with the tumor thickness obtained on histopathology. Furthermore, 9 of the 19 patients had tumors that neither reader was able to identify on CT, whereas all 19 OTSCC tumors were able to be identified on US.

Prior research has been performed on this topic noting that tumor thickness as assessed on US demonstrated a superior correlation with histopathologic tumor thickness compared to MRI [3,4]. Weimar et al. showed that tumor thickness as measured on CT and MRI performed similarly in predicting histopathologic tumor thickness [5]. They also demonstrated a higher degree of correlation between CT- or MRI-measured tumor thickness and histopathologic tumor thickness (r = 0.76–0.80) than our study. One possible explanation for the discrepant results is the differences in tumor size within the cohorts. The tumors in the Weimar study were overall larger in size with more than half of the cases with tumors greater than 10 mm in thickness with a mean tumor thickness of 11 mm. In comparison, only 2/11 (18%) patients in our cohort had tumors greater than 10 mm in thickness by histopathology with a mean tumor thickness of 7 mm. It appears that the larger tumors are more amenable to visualization on cross-sectional imaging, which correlates with prior studies that demonstrated that MR and CT are less accurate for assessment of lesions smaller than 5 mm in thickness [6].

The reliable detection of OTSCC on US and strong correlation of tumor thickness assessed on intraoperative US with histopathology prove extremely promising to support US as a powerful tool for preoperative evaluation of patients with OTSCC. This improved characterization of tumor thickness may translate to improvement in treatment planning and patient care. Accurate visualization and depiction of tumor shape, contour, and deep margin on US allows for better planning of surgical resection trajectory, in order to achieve a balance between resecting enough tissue for clear margins, and not so much as to compromise tongue function unnecessarily. Since neck dissection is recommended in patients with OTSCC tumors with a thickness of 4 mm or greater [7,8], accurate determination of tumor extent on US pre-operatively allows the surgeon to determine the need for up-front neck dissection a priori, which can then be performed in a clinically N0 neck during the same surgical episode as the tongue tumor resection.

There are, in particular, several limitations for CT that do not pose obstacles for US. It is extremely difficult, if possible at all, to determine the DOI of an OTSCC on CT due to close apposition of the tongue and tumor with adjacent structures such as the buccal surfaces, palate, and teeth, while both DOI and TT can be feasibly and accurately measured on US [2]. This is an important point since DOI is now considered by the AJCC as a key measurement for determination of the T category in tumor staging. Metallic dental artifact, which are not uncommon among adults of increasing age, do not negatively affect image visibility and diagnostic accuracy of an US as it does a CT. US can also be performed targeting the exact site of clinical concern, while it can be difficult to know the area of concern on a CT during image interpretation, especially for the tongue where placement of a radiopaque marker is not logistically feasible.

There are several limitations to this study. The study cohort is relatively small to date, because not all patients with OTSCC at our institution underwent CECT and intraoperative US with resection by the same surgeon. The intraoperative US had the advantage in that the studies were performed while being able to visualize the lesion grossly. However, the CTs were interpreted with knowledge of tumor laterality, and even for the tumors that were largest and among the most easily visible on CT, diagnostic capability was low, with the mean diagnostic confidence scores less than a value of 1 on the Likert scale, in between not confident and less confident. Lastly, all the intraoperative US in this study were performed with the patients under general anesthesia. Pre-operative US examination performed in an outpatient setting without anesthesia may face challenges such as patient discomfort and motion. That said, we have performed diagnostic intraoral US in an outpatient setting for tongue lesions with success, but further studies will be beneficial to determine the feasibility and accuracy of US evaluation of OTSCC in awake patients.

5. Conclusions

In this study, preoperative CT was found to be neither accurate nor reliable for assessment of primary OTSCC tumor thickness, while readily visible and well-defined on US. cTT has poor correlation with hTT while uTT has excellent correlation with hTT. Taken together, the results suggest a potential updated algorithm in the imaging work-up and preoperative planning of an OTSCC, which incorporates US in addition to CT in the pre-operative assessment pathway. Each plays a unique and important but different role, with US allowing accurate and reliable assessment of local tumor extent, thereby aiding tumor resection trajectory planning and determination of the need for up-front neck dissection in a clinically N0 neck, and cross-sectional imaging allowing optimal evaluation of the extent of invasion into adjacent structures and of overall nodal status.